Microscopic mechanisms for the Fermi-liquid behavior of Nb-doped strontium titanate

Microscopic mechanisms for the Fermi-Uquid behavior of
Nb-doped strontium titanate

S. N. Klimin" and J. Tempere''
Theorie van Kwantumsystemen en Complexe Systemen (TQC),
Universiteit Antwerpen, Universiteitsplein 1, B-2610 Antwerpen, Belgium

D. van der Marel
Departement de Physique de la Matiere Condensee,
Universite de Geneve, CH-1211 Geneve 4, Switzerland

J. T. Devreese

Theorie van Kwantumsystemen en Complexe Systemen (TQG),
Universiteit Antwerpen, Universiteitsplein 1, B-2610 Antwerpen, Belgium

(Dated: March 6, 2013)

Abstract

The relaxation rate in Nb-doped strontium titanate involving different scattering channels is
investigated theoretically. It is demonstrated that the total relaxation rate in SrTii_^Nbx.03 is
provided mainly by two mechanisms. The Baber electron-electron scattering with participation of
both Coulomb and phonon-mediated electron-electron interactions provides the T^-dependence of
the relaxation rate. The scattering on the potential landscape caused by impurities is responsible
for the residual relaxation rate at low temperatures. A good agreement with experiment is achieved
accounting for all phonon branches in strontium titanate, both the optical and acoustic phonons.
It is shown that the effective electron-electron interaction can be attractive in strontium titanate,
and provides superconductivity at low temperatures and Fermi-liquid response in a wide range
of temperatures. Thus our microscopic model supports the notion that superconductivity and
Fermi-liquid properties of n-type SrTiOs have a common origin.

1

I. INTRODUCTION

In Ref. j^, a common origin of the Fermi-liquid properties and superconductivity in
Nb-doped strontium titanate has been postulated and induced from the phenomenological
treatment of the resistivity data and superconducting critical temperatures T^. In the present
work, we deduce the temperature dependence of the resistivity and the superconducting Tc's
from microscopic model calculations. The relaxation rate in SrTii_^.Nb^.03 can be accounted
for by various scattering mechanisms. The Fermi-liquid properties of a charge carrier gas
in a crystal are provided by the effective electron-electron interaction. This interaction is
an interplay of the Coulomb repulsion and the phonon-mediated attraction provided by
the Frohlich electron-phonon interaction with the optical phonons. Strontium titanate is a
strongly polar crystal with static dielectric constant ~ 240 and high-frequency dielectric
constant £00 ~ 5.44 [4]. The ratio rj = Eoo/so is very small in strontium titanate. Due to
a small rj, the Coulomb repulsion and the optical phonon-mediated attraction significantly
compensate each other. Thus the contribution of the other electron-phonon interactions,
e.g., the deformation interaction with the acoustic phonons, can be important for the sign of
the total effective electron-electron interaction. The fact that n-doped SrTiOa is a supercon-
ductor implies that in this material the phonon-mediated attraction due to both optical
and acoustic phonons can overcome the Coulomb repulsion. This yields superconductivity
in strontium titanate at low temperatures about T ~ 1 K. In this connection, we verify the
suggestion ^ that the DC resistivity and the relaxation rate in Nb doped SrTiOs can be due
to the effective electron-electron interaction. There are two channels for the Fermi-liquid

response of metals and strongly doped semiconductors with conip

7-

ex conductivity bands due
9| scattering processes.

to the electron-electron interaction: the normal and Umklapp

Several other scattering mechanisms besides the effective electron-electron interaction are
present in strontium titanate. For completeness, they must be included in the treatment,
as far as they may contribute to the total resistivity. In the present work, the following
additional scattering mechanisms are considered.

(1) The electron-phonon interaction can contribute to the effective electron-electron scat-
tering as mentioned above. Besides this, the direct scattering of the electrons by the LO
phonons can contribute to the resistivity of the polar crystals in the same way as to the
optical absorption 0, 10, 11 1.

2

(2) In the context of small polarons, the electron-phonon scatterin g re sults in a tem-
perature dependence of the resistivity which is close to the T^-behavior 12]. Therefore the
electron - LO-phonon scattering may influence the Fermi-liquid-like temperature dependence
of the DC response of the charge carriers.

As established in Refs. {4, 10|, the polarons in strontium titanate are large (Frohlich)

polarons rather than the small polarons treated in Ref . [12| . In this connection, the electron
- LO-phonon scattering is considered here in terms of the large polarons.

(3) Finally, the scattering of the electrons on the potential landscape induced by the
impurities (for example, the Nb donors) can bring a non-negligible contribution to the total
relaxation rate and resistivity, especially in the low-temperature regime.

II. THEORY AND RESULTS
A. Baber scattering

Within a parabolic model for a simple conductivity band, the electron-electron inter-
action due to the normal (i.e., within the first Brillouin zone) scattering processes does
not contribute to the carrier electric response owing to momentum conservation, even for an

anisotropic band 13|. However, for a non-parabolic and/or complex conductivity band with.

e.g., light and heavy carriers, the carrier response can be non-zero as first found by Baber

14

15| . Thus the resistivity can be provided by the interaction between charge carriers
with different band masses or between electrons and holes.

The Baber scattering mechanism is the most promising candidate to provide the main
contribution to the total relaxation rate of Nb doped strontium titanate. We calculate
;he resistivity using the Boltzmann equation. Further on, we follow the notations of Ref.

161] . The non-equilibrium distribution function /„^k for the carriers in the n-th subband is

determined as

where f^l = f (^n.k) is the equilibrium electron distribution function in the n-th subband
of the conduction band,

fi^nM) = ^^ , (2)

e "bt +1

3

and e„,k is the electron energy.

The function $„^k is a measure of the deviation from equihbrium in the electron dis-
tribution. The inner product of two real functions and $ is defined by (\1',$) =
J dh '^n,k^n,k- The Boltzmann equation can be written as

X = P^, (3)

where P is the scattering operator which transforms the function $ into another function
= P$, and X represents the left-hand side of the Boltzmann equation,

X = -v„,k ■ ^^VT - v„,k • e^^E. (4)

Here, v„ k is the velocity in the n-th subband.

The collision integral of the Boltzmann equation can be represented as ($,P$). The
Boltzmann equation in the form Q implies that

($,X) = ($,P<I>). (6)

The variational principle established by Ziman \1G\ states that the solution of the Boltzmann
equation gives ($, P$) its maximum value. Thus $„^k can be approximated by a trial
variational function.

In the notations of Ref. the DC resistivity is expressed by the formula

p = ^ ' / f7)

mx{E = i))f

where X {E = 1) represents the left-hand side of the Boltzmann equation in a unit electric
field E = 1 (and in the absence of the temperature gradients). The normalization factor is
determined by the expression

($,X(E=l)) = 2 5:/ev.,k$„,k^^-^^ (8)

The collision integral for the electron-electron scattering processes is

, ^ „ ^ V 1 f dki (iko dk-i dkA , , ^ ^ ^ x 2

(*. - ^ E y p;,. (i^ (i^ (*- + - - *-*)^

xV{n,l,n',2^n,3,n',A). (9)

4

Here, V {n, 1, n', 2 — )■ n, 3, n', 4) is the scattering probability,

V (n, 1, n', 2 ^ n, 3, n', 4) = ' / (e.^kj / {Sn'^) [1 - / (^n.ks)] [1 - / (^n',kj]

271

X ^ " l^n,ki "I" ^n',k2 ^n,k3 ^n',k4j
X (27rf 5(ki+k2-k3-k4),

with f^k^'^kg the matrix element of the effective electron-electron interaction, which includes
both the Coulomb repulsion and the phonon-mediated attraction.

n

The trial variational function $„ k is chosen in the form pj|

$n,k = v„_k ■ u

where u is the unit vector parallel to the applied electric field. Thus the collision integral is
1 1 27r

($,P$)

2kBT{2ny h

^ j dkidk2dksdki ((v„,,ki + ^n'M - v„,k3 - v„',k4) ■ u)

X {Ut^iyf (^n,kj / (en',kj [1 - / (e„,k3)] [1 - / (^n',kj]
X 6 (£„,ki + £n'M2 - ^n,k3 - ^u'm) ^ (kl + ks - kg - k4) .

(10)

The electrons in the conduction band are described by the matrix Hamiltonian from Ref.

H = 4

^ ei{k) ^

62 (k)
^ ssik) ^

with the energies

El = ts sin^

£2 = sin

£3 = tn sin

) + sm '

2

+ ts sin
+ sin

2

2 / OjQ^y

2

2 / (^O^y

2 J " V 2
where oq is the lattice constant. The matrix W

(

W =

\

2D ^ ^

+ t,r sin
+ t,r sin
+ ^5 sin

\
/

2

2 / (^0^2

2

2 / ciokz

(11)

(12)

(13)

describes the mixing of subbands within the conductivity band. We use the values of the
band parameters ts,t.,,D,C, from Ref. [3|: ts = 35 meV, t,r = 615 meV, C, = 18.8 meV and
D = 2.2 meV.

In the present treatment, we neglect the band mixing because it is relatively small, and
assume the anisotropic parabolic dispersion in each of three subbands of the conductivity

band fas in Refs. 13

with the tensor effective masses

m

in)

The masses m^"'' can take two values: the "light" mass

(j = x,y,z, n = 1,2,3).

and the "heavy" mass

= -^Ti-- Using the above band parameters of strontium titanate and the lattice constant
ao = 3.905 A, we arrive at the band masses m.„ ~ 1.6me and ~ 29me, where me is the
electron mass in vacuum.

The effective electron-electron interaction in SrTiOa is considered in the following way.
In the standard diagram technique, the effective interaction corresponding to an elementary
electron-electron scattering process for a single-branch electron-phonon system, is given by

the expression

u,

k,k'

f/i5/^ (q,a;)

Uc {<D + V^'^ (q,

CO

where V^^^ (q, to) is the free-phonon Green's function,

2ljJr

2)(o)

9 9 '

(14)

q is the momentum transfer q = k' — k, and u is equal to the transition frequency u =
(£n',k' — £n,\s.) /fi (where n,n' are the indices of the subbands of the conduction band). Thus
the effective interaction is

Uti^^ (q,u;) = f/c(q)-|Fq

2lTlUr,

\2 •

(15)

The effective interaction f[T5|) is written for a one-branch phonon system and does not account
for screening. For our calculations, both the screening and the multi-branch phonon system
must be accounted for. The extension of the above formulae to the multi-branch phonon
system is straightforward:

(ac) T r(ac) ^

2l^q,A |^q,A|'

(eff)

,q,wj

C

1 2ut^

UJ.

q,A

h ( {ac)

Ulc

(16)

6

where A labels different optical-phonon branches. The contribution to the effective interac-
tion due to the acoustic phonons (with the interaction amplitudes Vq""^^) is written here as
a separate term. The screening for the Coulomb interaction and for the phonon-mediated
interaction can be introduced here as in Ref. Ilj

rjieff) . . ^ 4vre- _ 1 \V^,,\' 2a;,,. _ 1 ^^^^

yiac)

where eg (q, u) is the electron screening factor. Note that screening is different for the
Coulomb and phonon-mediated interactions, and that the contribution due to the acoustic
phonons is not screened.

In the experimental situation of Ref. jsl, the Fermi energy of the electrons in the con-
duction band substantially exceeds their thermal energy ksT. Therefore only the electrons
in a thermal layer near the Fermi surface bring a dominating contribution to the relaxation
rate. In other words, the energies of the relevant electrons are close to the Fermi energy,
and the difference En'y — £n,iL ~ ksT is relatively small with respect to the Fermi energy.

The major contribution to the collision integral fIlOp lies in the range of the momentum
transfer q oc 2kp, where kp is the electron wave vector at the Fermi surface (in general,
angle- dependent due to the band anisotropy). In SrTiOa, the corresponding acoustic-phonon
energy satisfies the condition Hu^^^ > ksT in the range T < 100 K for all samples treated in
the experiment sl except the lowest-doped sample, for which hoj^^^ /ksT ~ 1 at T = 100 K.
For the lower temperatures and/or for the higher doping contents the aforesaid condition is
fulfilled. The LO-phonon energies in SrTiOs are much higher than the thermal energy in the
considered temperature range. Thus, owing to the fact that the electron-electron scattering
effectively occurs close to the Fermi surface, retardation effects in the phonon-mediated
interaction can be neglected. Therefore we can approximate the effective interaction by the
expression

T/(«c)

g%ooee (q) ^ ^q.A [te (q)]' hw^^ '
where eg (q) is the static electron screening factor. The different screening of the Coulomb
and phonon-mediated interactions, however, can strongly influence the resulting interaction.
In the present calculation, the Thomas-Fermi (TF) screening factor is used in the effective

7

interaction:

ee (q) = 1 +

where is the inverse TF screening length,

1/2

(19)

(20)

uq is the carrier density, and Ep is the Fermi energy. Note that eg (q) is the static screening
factor for the electron gas (without the lattice polarization). Therefore Kg contains the high-
frequency dielectric constant rather than the static dielectric constant ^o- The material
parameters for the Frohlich interaction with the optical phonons are taken the same as in
Ref. {4]. The values of the TO- and LO- phonon frequencies, the actual electron densities

and the plasma frequencies Up determined from the experiment
1.

10| are represented in Table

TABLE I: Optical-phonon frequencies, electron densities and plasma frequencies of doped strontium
titanate

X

X = 0.1%

X = 0.1%

X = 0.2%

X = 0.2%

X = 0.9%

X = 0.9%

x = 2%

X = 2%

T

T = 7 K

r = 300 K

T = 7 K

r = 300 K

T = 7 K

T = 300 K

T = 7 K

T = 300 K

fvjjTO,! (meV)

2.27

11.5

2.63

11.5

6.01

12.1

8.51

13.0

hwLO,i (meV)

21.2

21.2

21.2

21.2

21.2

21.2

21.2

21.2

^TO,2 (meV)

21.2

21.8

21.2

21.8

21.2

21.8

21.2

21.8

^LO,2 (meV)

58.4

58.4

58.4

58.4

58.4

58.4

58.4

58.4

fvjjTOfi (meV)

67.6

67.1

67.6

67.1

67.6

67.1

67.6

67.1

^LO,3 (meV)

98.7

98.7

98.7

98.7

98.7

98.7

98.7

98.7

no (cm~^)

1.7 X 10^9

1.7 X 10^9

3.4 X 10^^

3.4 X 10^^

1.5 X 10^0

1.5 X 10^0

3.4 X 10^0

3.4 X 10^0

hjjjp (eV)

0.1

0.1

0.14

0.14

0.29

0.29

0.44

0.44

For the acoustic-phonon contribution, we use the interaction amplitudes for the deforma-
tion potential from Ref. 18 1

rriD

with the dimensionless couphng constant

On

(21)

(22)

8

where n is the density, Ed is the deformation potential, v is the sound velocity, and m/j is the
density-of-state band mass. It is i.a. noting that the acoustic-phonon interaction amplitude
(PTj) in fact does not depend on the mass mD- The sound velocity in strontium titanate is

taken f ^ 8.1 x lO^ms [19 1.

The relaxation rate is determined using its relation to the resistivity.

(23)

where Up is the plasma frequency for the electron gas in the conductivity band determined
in the same way as in Ref. js], and listed in Table 1. In Fig. 1, we represent the numerical
results for the relaxation rate provided by Baber scattering as a function of temperature
for the actual electron densities, which are determined from the experimental values of the

n

doping level given in Ref. m. We possess the measured values for all material parameters of
strontium titanate, except the deformation potential E^. In this connection, the value E^, =
23.3 eV has been chosen to fit the T^-dependence of the relaxation rate to the experimental
data for the lowest doping level x = 0.1%. For the other densities, we keep one and the
same value of E^,. Typical values of the deformation potential in other crystals found in the
literature are about 10 to 30 eV, so that the chosen value is realistic.

The relaxation rates calculated with the same value of Ed for all densities reproduce
the density dependence of the experimentally obtained relaxation rate fairly well, except a
residual background contribution which does not vanish when the temperature tends to zero.

Therefore the suggestion j^l that the T^-dependence of the DC resistivity in SrTii_a;Nba;0

3

can be provided by the Baber scattering mechanism is supported by the present calculation.

When both optical and acoustic phonons are taken into account, the effective electron-
electron interaction at the Fermi surface can become attractive. With optical phonons only
(i.e., neglecting acoustic-phonon contribution) this attraction is not possible. Moreover,
the effective electron-electron interaction without participation of the acoustic phonons is
approximately Coulomb-like. The relaxation rate for a Coulomb-like electron-electron in-
teraction monotonously decreases with an increasing density. The density dependence of
the relaxation rate observed experimentally can be explained by a relative increase of the
attraction provided by the acoustic phonons with respect to the Coulomb repulsion and the
interaction due to the optical phonons. The latter two become more strongly screened when
the density rises, while the deformation potential is not screened. The increasing density de-

pendence of the relaxation rate therefore supports our hypothesis that the phonon-mediated
attraction (provided by both optical and acoustic phonons) in strontium titanate can over-
come the Coulomb repulsion even in the normal phase - at sufficiently high temperatures,
when superconductivity does not exist.

B. Other mechanisms

We suggest that the relatively small residual relaxation rate in the Nb-doped SrTiOs
can be accounted for by the scattering of the electrons on the potential landscape created
by the impurities. The contribution to the DC resistivity due to the scattering of the
electrons by the potential of the impurities is calculated using the approximation of the
time-dependent relaxation time within the Boltzmann equation approach. Assuming the
impurities chaotically distributed in space, the DC conductivity tensor is determined by the
expression (see Ref. |20t])

dk f dfo (£:„.k)

3^n (k)

(27r

den,]

(24)

where (v„ k)^. are the components of the electron velocity in the conduction band.

For the relaxation time T„(k), we apply the approximation of the isotropic three-
fold degenerate conduction band with the effective density-of-states band mass mo =
m'":^] m^Jl^^ . In this approximation, (k) = r(k), and the conductivity tensor

is reduced to the scalar expression a

1 _
P

dk

(2^

■A;V(k)

dfo je]
de

(25)

e=e{k)

where p is the DC resistivity. The momentum-dependent relaxation time r (k) for the
chaotically distributed impurities with the density rij is determined using the differential
scattering cross-section a {k, 6) as

1

- = 27fniVk

'0

r

(k)

2nniVk / a {k,9) {1 — COS!
Jo

de

(26)

The scattering cross-section a [k, 6) is calculated here in the Born approximation 21 |.

a{k,e)

m

D

47r2n4

U (r) e

dr

(27)

10

where U (r) is the potential created by an impurity, and q = k' — k is the momentum transfer
(with k' = k, because the scattering is elastic).

To the best of our knowledge, the true potentials created by the impurities in doped
strontium titanate are not yet reliably known. We suggest that the majority contribution
to the impurity scattering is due to the ionized niobium donors. Thus we assume that the
density nj ^ no (where no is the electron density), and that the potential created by an
ion can be modeled by a pseudopotential corresponding to the Yukawa (screened Coulomb)
potential:

U{v) = -^exp{-Kjr), (28)

where Uq = e'^/eo, ki is the inverse screening length for the impurity potential. The scat-
tering cross-section a [k, 0) with the potential (125]) is

a {k, 6) = ( '-^) ' (29)

Integrating in (126|) with (129!) . the inverse momentum-dependent relaxation time is obtained
to be

1 _ 27r^rajt;k / 2mz)[/o y 1
r{k)~ Kj \ f9 ) (4P + «:2)3/2- i )

The contribution to the total relaxation rate (125|) provided by the impurity scattering, l/r^^p
with f l30|) is calculated numerically.

Here, the parameter to be estimated is the inverse screening length kj. First, we can
calculate ki using the Thomas-Fermi (TF) approximation,

.r)=f^V'', (31)

where Ep = (vr^no)^^^ is the Fermi energy for the threefold degenerate conduction band.
We use here the static (rather than high-frequency) dielectric constant because the ions do
not move, and hence the lattice polarization (provided by the polar LO phonons) reduces
the Coulomb potential by the ratio Eoo/eo-

The results for the ^/^eff with (IT^ are shown in Fig. 2 (a). The values of
the zero-temperature limit are in fair agreement with the experiment jsl for the samples
with two higher doping levels x = 0.9% and x = 2%. However, for the two weaker doped
samples, the residual relaxation rate calculated using (l3T]l is overestimated with respect to
the experiment. Therefore we can suggest that the spatial cutoff of the impurity potential

11

can be provided also by additional mechanisms which can be irrelevant to the presence of the
electron gas and which therefore do not vanish at low electron densities. In this connection,
the contribution to the total relaxation rate due to the impurity scattering has been also
calculated using the trial screening radius r/ = ^ 0.76 nm which is one and the same
for all samples and close to k^'^^'' for the highest doping. The results for 1 / r^^^ in this case
are shown in Fig. 2 (6). We see that in this case, the residual relaxation rate is in good
agreement with the background for the total relaxation rate measured in the experiment j^.
Moreover, the present result adequately reproduces the experimental density dependence of
the residual relaxation rate without varying k/.

In our treatment of the impurity screening length, we suppose this quantity to be inde-
pendent of doping in order to arrive at the experimentally observed trend. The assumption
that /€/ does not depend on the doping means that the dominant part of the impurity poten-
tial for the scattering processes is essentially short-range. (Note that r/ 0.76 nm is about
twice the lattice parameter.) The Nb-ion is larger than the Ti-atom and as a consequence
the ions around the Nb-donors are radially displaced. This has consequences for the local
electronic structure and - as far as we know - nobody has analyzed this theoretically for
Nb-donors in SrTiOa. Empirically we know that even for a low Nb-concentration, no bound
donor-states are formed. This is totally different from the case of Si doped with phosphor.

For the Yukawa potential fl28|) . a critical value n'f' exists such that for ki > K^f\ there
are no bound states in that potential. The Monte Carlo calculation for a particle with an

22I 1 gives

isotropic parabolic dispersion law

R^"'^ ^ 1.1906?/, (32)
where k^^^ and U are, respectively, n^f^ and U in the atomic units. We can estimate k\^'^ for an

~ (c)

anisotropic band using the parameters k) and U expressed in the effective atomic units for
a crystal: kf^ = K'f'a*^, U = U / {a*QEl), with the effective Bohr radius a*^ = £0^^/ (^d^^)
and the effective Hartree energy = m^e^ (^o^)^- In Nb-doped strontium titanate, the
critical inverse screening radius corresponding to (!32|) . is rj^^ ^ 2.5 nm. The aforesaid value

~ ~ ~(c)

rj ^ 0.76 nm yields kj ^ 3.9U, substantially larger than k) . Thus, in accordance with the
aforesaid experimental evidence, there are no bound states in the Yukawa potential with the
parameters of the Nb-doped SrTiOs chosen in the present calculation.

The other mechanisms which might contribute to the resistivity of SrTii.^^Nb^^Os are the

12

direct scattering of the electrons by the optical phonons and the Umklapp electron-electron
scattering. These contributions have been separately calculated within the same kinetic
equation approach as the aforesaid contributions due to the Baber and impurity scattering.
In Fig. 3, the contribution to the relaxation rate due to the electron - LO-phonon scattering
is plotted as a function of the temperature for the actual electron densities in the samples
studied experimentally in Ref. Under the experimental conditions of Ref. js], the elec-
tron - LO-phonon scattering contribution to the relaxation rate appears to be relatively
small with respect to the relaxation rate provided by the Baber mechanism. Furthermore,
the relaxation rate due to the electron - LO-phonon scattering strongly decreases with de-
creasing temperature. Therefore the electron-phonon scattering is not one of the dominating
mechanisms of the DC conductivity in Nb doped SrTiOs.

Finally, the Umklapp electron-electron scattering was considered in Ref. j^l as one of
possible sources for the DC resistivity in Nb-doped SrTiOa. For the electron resistivity
in metals, the Baber mechanism plays a minor role, and the Umklapp scattering can be
sufficient to explain the Fermi-liquid temperature dependence of the resistivity

[JS- How-
ever, the Umklapp electron-electron scattering remarkably contributes to the resistivity only
when both the initial and final momenta of the two electrons in the elementary scattering
process lie in the thermal layer near the Fermi surface. Therefore, e.g. in the simple case
of a spherically symmetric conductivity band, the Umklapp scattering does not contribute
to the resistivity when the maximal possible value of the momentum transfer gmax =
(where kp is the Fermi wave vector) is smaller than the modulus of the reciprocal lattice
vector g. Therefore the Umklapp processes can be non- negligible only at sufficiently high
densities. For an anisotropic conductivity band with a warped Fermi surface (that is just
the case in SrTiOs) the restriction for the density can be softened with respect to that for a
spherically symmetric band. However, it is not a priori known whether the Umklapp scat-
tering is relevant for the carrier densities in the experiment j3] . Therefore we have estimated
the Umklapp contribution to the relaxation rate in strontium titanate within the kinetic
equation approach.

In Fig. 4, the relaxation rate provided by the Umklapp electron-electron scattering
(multiplie d by h/Q"^, where ^ < 1 is the dimensionless interference factor for the Umklapp

processes

161]) is plotted as a function of the electron density for the temperature T = 80

K. The relaxation rate provided by the Umklapp processes in strontium titanate is not

13

vanishingly small for densities higher than the threshold value no ~ 8 x 10^" cm~^. However,
the carrier densities achieved in the experiment sl are substantially lower than this threshold
value. In the figure, the arrows show the actual electron densities in the experiment j^. For
the experimental densities uq ~ 10^^ cm^^ to 10^° cm~^, the relaxation rate provided by
the Umklapp electron-electron scattering is vanishingly small with respect to the measured
values of 1/r. The relaxation rate due to the Umklapp processes becomes non-negligible
in strontium titanate only for the electron densities no > lO^^cm"^, much higher than
the densities relevant for the experiment j^. The conclusion follows that the contribution
by Umklapp processes to the Fermi-liquid behavior of relaxation rate and resistivity in
SrTii_a,Nba;03 is negligible. In summary, the electron - LO-phonon and Umklapp scattering
contributions to the DC resistivity of the Nb doped strontium titanate bring only relatively
small corrections in the present treatment as far as it is related to the interpretation of the
experiment

The relative contributions of different scattering mechanisms to the total relaxation rate
as a function of temperature are shown in Fig. 5 for the sample with the doping content x =
0.002. In the low-temperature range, the residual scattering of the electrons on the potential
landscape of the impurities dominates. For higher temperatures, the Baber scattering plays a
key role providing the Fermi-liquid T^-dependence of the total relaxation rate. The electron
- LO-phonon scattering only slightly contributes to the total relaxation rate and only at
sufficiently high temperatures. Thus the main contribution to the relaxation rate is brought
by the Baber scattering at moderate temperatures and by the scattering by the donors at
low temperatures.

The total relaxation rate calculated taking into account the Baber, electron-impurity,
and electron - LO-phonon scattering mechanisms is plotted in Fig. 6. The temperature
dependence of the relaxation rate is completely due to the Baber mechanism. The scattering
by donors is responsible for the residual relaxation rate which constitutes the background
resistivity in SrTii_a;Nba;03. We see that the aforesaid mechanisms convincingly explain
both the temperature and density dependences for the experimentally measured relaxation
rate.

14

III. CONCLUSIONS

In conclusion, the DC resistivity of Nb-doped strontium titanate is explained in terms
of two dominating mechanisms: Baber scattering with participation of both Coulomb and
phonon-mediated electron-electron interactions provides the T^-dependence of the resistiv-
ity and of the relaxation rate, while the scattering on the potential landscape caused by
impurities yields the residual relaxation rate which does not vanish at T = 0. The calcu-
lated relaxation rates are in a good agreement with the experiment ^ . Thus the hypothesis
on a common origin of two phenomena in SrTiOS - superconductivity and the Fermi-liquid
behavior of the resistivity - is supported by the microscopic calculations.

Acknowledgments

This work was supported by FWO-V projects G.0356.06, G.0370.09N, G.0180.09N,
G.0365.08, the WOG WO.035.04N (Belgium), the SNSF through Grant No. 200020-140761
and the National Center of Competence in Research (NCCR) "Materials with Novel Elec-
tronic Properties-MaNEP" .

[a] On leave of absence from: Department of Theoretical Physics, State University of Moldova,
str. A. Mateevici 60, MD-2009 Kishinev, Republic of Moldova.

[b] Also at Lyman Laboratory of Physics, Harvard University, Cambridge, MA 02138, USA.
[3] D. van der Marel, J. L. M. van Mechelen, and I. I. Mazin, Phys. Rev. B 84, 205111 (2011).
[4] J. T. Devreese, S. N. Klimin, J. L. M. van Mechelen, and D. van der Marel, Phys. Rev. B 81,

125119 (2010).

[5] C. S. Koonce, M. L. Cohen, J. F. Schooley, W. R. Hosier, and E. R. Pfeiffer, Phys. Rev. 163,
380 (1967).

[6] N. Reyren, S. Thiel, A. D. Caviglia, L. F. Kourkoutis, G. Hammerl, C. Richter, C. W. Schnei-
der, T. Kopp, A.-S. Ruetschi, D. Jaccard, M. Gabay, D. A. Muller, J.-M. Triscone, and J.
Mannhart, Science 317, 1196 (2007).

[7] W. E. Lawrence and J. W. Wilkins, Phys. Rev. 7, 2317 (1973)

[8] A. H. MacDonald, Phys. Rev. Lett. 44, 489 (1980).

15

[9] A. H. MacDonald, R. Taylor, and D. J. W. Geldart, Phys. Rev. B 23, 2718 (1981).

[10] J. L. M. van Mechelen, D. van der Marel, C. Grimaldi, A. B. Kuzmenko, N. P. Armitage, N.

Reyren, H. Hagemann, and I. I. Mazin, Phys. Rev. Lett. 100, 226403 (2008).

[11] J. Tcmpcrc and J. T. Devreese, Phys. Rev. B 64, 104504 (2001).

[12] G. M. Zhao, V. Smolyaninova, W. Prelher, and H. Keller, Phys. Rev. Lett. 84, 6086 (2000).

[13] H. Maebashi and H. Fukuyama, J. Phys. Soc. Jap. 66, 3577 (1997).

[14] W. G. Baber, Proc. R. Soc. A 158, 383 (1937).

[15] T. Giamarchi and B. S. Shastry, Phys. Rev. B 46, 5528 (1992).

[16] J. M. Ziman, Electrons and Phonons (Oxford U.P., Oxford, England, 1960).

[17] G. D. Mahan, Many-Particle Physics, second edition (Plenum Press, 1990).

[18] F. M. Peeters and J. T. Devreese, Phys. Rev. B 32, 3515 (1985).

[19] R. O. Bell and G. Rupprecht, Phys. Rev. 129, 90 (1963).

[20] A. I. Anselm, Introduction to Semiconductor Theory (Mir, Moscow/Prentice-Hall, Englewood
Cliffs, NJ, 1981).

[21] L. D. Landau and L. M. Lifshitz, Quantum Mechanics Non-Relativistic Theory, Vol. 3.

[22] Y. Li, X. Luo and H. Kroger, Sci. China G: Phys. Mech. and Astr. 49, 60 (2006).

16

1 ' 1 ' 1 ' r

T(K)

FIG. 1: Relaxation rate provided by the Baber scattering in Nb-doped SrTiOa as a function of
the temperature for the actual electron densities corresponding to the experimental values of the
doping [3|].

17

T(K)

FIG. 2: Residual relaxation rate provided by the scattering of the electrons on the impurities in
Nb-doped SrTiOs as a function of the temperature (a) using the Thomas-Fermi screening length,
(b) using the model screening length, the same for all samples.

18

0.3

0.2 -

0.1 -

0.0

1 1 1 1 1 1 1
Electron - LO-phonon scattering

1

1

f
/

f

f

//

x = 0.001

f •

//

x = 0.002

// -

x = 0.009

•/
1:

x = 0.02

oOt^ 1

!! /
// /

/'-' / y

1

20

40

60

80

r(K)

FIG. 3: Contribution to the relaxation rate in Nb-doped SrTiOs due to the direct electron
LO-phonon scattering as a function of the temperature for the electron densities from Ref. Q].

19

0.4

0.3 -

>

0.2 -

0.1 -

0.0

— ' 1 ' r

r=80K

-//-

1 — ' — I — ' — I — ' — I — ' — r

Contribution
due to the Umklapp
processes

X- 0.001

X = 0.002

jc = 0.009

X = 0.02

0.0 0.1 0.2 0.3

0.5 1.0 1.5 2.0 2.5 3.0

21 -3

Density (x 10 cm )

FIG. 4: Relaxation rate due to the Umklapp electron-electron scattering in SrTii-ajNb^Oa multi-
plied by h/Q"^ (where G is the interference factor for the Umklapp processes {igI ) as a function of
the electron density for T = 80 K. The arrows indicate the electron density for the samples of the
experiment Q]. There are different scales at the x axis for the densities no < 4 x lO^'^cm"^ and
no > 4 X 10^° cm~^.

20

T{K)

FIG. 5: Relative contributions of different scattering mechanisms to tlie total relaxation rate as a
function of temperature for the doping content x = 0.002.

21

12

10 -

8 -

>

4

2 -

Data Theory

o x = 0.001

A x = 0.002

V x = 0.009

O x = 0.02

/7
'O _

/I

• O

FIG. 6: Curves: the calculated total relaxation rate in Nb-doped SrTiOs as a function of the
temperature for the actual electron densities corresponding to the experimental values of the doping
3|. Symbols: experimentally determined relaxation rate in Nb-doped SrTiOs from Fig. 2 of Ref.

i

22